Parameter Entities for Content Models in P2 draft of 20 Oct 92, CMSMcQ, rev. 23 Oct TEI-internal distribution only 1 Genre-specific classes: these are sets of CHUNK-level and INTER-level elements which can occur only in individual genres. (N.B. any element must either be common to all genres or unique to a single genre. No elements common to some subset of genres may be defined.) Class names: prose, verse, drama, spoken, letters, dictionaries, terminology, collections. Parameter entity references will take the form: %m.prose; %m.verse; %m.drama; etc. User-defined extensions can be included in the classes %x.prose;, %x.verse;, etc. 2 Other classes and parameter entities needed for Swiss DTDs: These are: the class 'common' and the parameter entities 'seq', 'gen.seq', and 'mixed.seq'. common (%m.common): class of chunks or inter elements not marked for genre. User-defined additions can be included in the p.e. x.common. Definition will take the form: (The list of constituent elements is not final.) seq (%seq;): standard p.e. name for sequence of whatever chunks are legal in DIVs, as described in SWISS.DTD. Typical definitions will be etc. These definitions must occur in the genre-specific file included as the TEI base. The special Swiss and Anarchist bases will include appropriate definitions of seq, as described below. The parameter entity 'seq' corresponds to the 'sq' of SWISS.DTD and to the p.seq of P1. Users may modify seq by redefining it. gen.seq (%gen.seq;): standard p.e. name for 'Swiss sequence', will be used in front- and back-matter 'generic DIVs'. Definition is: The entity 'gen.seq' corresponds to SWISS.DTD's 'swisssq'. Users cannot extend gen.seq except by redefining it in its entirety. mixed.seq (%mixed.seq;): standard p.e. name for 'anarchist sequence', will be used for wild-and-woolly texts with mixed-genre divs. Definition: The entity 'mixed.seq' corresponds to the 'freesq' of SWISS.DTD. Users cannot extend 'mixed.seq' except by redefining it in its entirety. 3 Structurally distinct classes: These are: phrase, inter, and chunk. phrase: class of all low-level tags which can only occur within paragraphs or other chunks (those in old 'broth'), probably included indirectly through semantic-group classes (see below). Provisional definitions: Few if any tags in this class will be genre-specific; any genre-specific tags which do occur will simply be defined into the class. In normal cases, such genre-specific phrase-level tags won't cause problems in other genres because they won't have been defined. When Swiss Divs are used, they will have been defined and may cause undesirable looseness in chunks defined as taking '(#PCDATA | %m.phrase; | %m.inter;)*'. This possible undesired looseness is accepted as being preferable to defining genre-specific phrase classes which will almost never be used; we can add such classes later if it proves necessary. inter: class of all elements which can occur either within or between chunks. Provisional definitions: chunk: class of all elements which can only occur between chunks. (Alternate name: block.) The class 'chunk' corresponds to the class of paragraph-level elements and 'crystals' in P1. Provisional definitions: 4 Common Content Models These are sometimes mixtures of classes, not simple sequences, so their names do not all take the form XXX.seq. (Discussable.) The entity %phrase.seq corresponds to the %broth; of P1 and of most P2 drafts. In many cases, however, it should probably be replaced by %paraContent; since most phrase-level elements must be able to have notes (and possibly other inter-level elements) attached within them. The entity %chunk.seq corresponds to the %p.seq; of P1. The entity %paraContent corresponds to the %soup; (i.e. broth + chunks) of P1. Elements which contain either a series of chunks or possibly only a word or phrase (i.e. which either function like a single paragraph or else include a series of paragraphs) can be defined using the following model. Key members of this set are list items and notes. Some of these names are pluralized, others not. This may need to be normalized.